OPTIMIZING JOINS BETWEEN 2 PARTITIONED RELATIONS IN DISTRIBUTED DATABASES

This paper analyzes the problem of optimizing the execution of a join between two partitioned relations in a distributed database. The approach followed in this paper consists in developing a formal model of partitioned database and of distributed joins, and then discussing all the relevant optimization problems in this framework. These problems include the 'semantic optimization of joins, semi-join reduction, and the 'physical' optimization of joins. The 'physical' optimization of joins is formulated as a linear zero-one program, and then a theorem is presented which reduces the complexity of the program, Finally, an overall procedure for the minimization of join execution costs in distributed databases is developed, which uses the formulations of the above problems in sequence.